ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume X-4/W1-2022
https://doi.org/10.5194/isprs-annals-X-4-W1-2022-159-2023
https://doi.org/10.5194/isprs-annals-X-4-W1-2022-159-2023
13 Jan 2023
 | 13 Jan 2023

FOREST MODELING AND INVENTORY ESTIMATION USING LIDAR DATA

B. Farajelahi, F. F. Eya, and H. Arefi

Keywords: Individual Tree Segmentation, Statistical Information, Filtering, Classification, Point Cloud, Digital Surface Model (DSM), Digital Terrain Model (DTM), Canopy Height Model (CHM)

Abstract. Aerial laser scanners find rapidly growing interest in photogrammetry and remote sensing as an efficient tool for reliable three-dimensional extraction and modelling of forest inventory information. In addition to interactive measurements in 3D point clouds, techniques for automatic extraction of objects and determination of geometric parameters form a high and important research issues (Maas, Bienert et al. 2008). This paper presents a novel approach on the extraction and modelling of individual trees from the Idaho National Forest in the USA and calculation of the statistical estimation for each extracted segment for future analysing. Lidar point cloud contains three-dimensional structure information which is used to estimate the statistical information for each tree segments. In this study, we worked on the raster surface made directly from the LiDAR point cloud and two main models, namely the digital terrain model (DTM) and the digital surface model (DSM), are generated when the point clouds are processed by the filtering method. Then we have used the segmentation techniques to extract the tree segments which is a triggering process that facilitates the extraction of statistical information such as crown diameter, eccentricity, and other additional attributes. The proposed individual tree segmentation method results in 73% correctness, 92% completeness and 81% F1-score.